This paper attempts to show that it is possible to analyze a problem solving protocol in which the subject spontaneously generates a series of analogies. The qualitative physics problem given to the subject describes a situation where that which is unfamiliar is often solved by relating it to several analogous situations that are more familiar. Also described as important in creative problem solving are generating Gedanken experiments, generalizing and specializing, and generating extreme cases. This analysis suggests that three processes are essential in reasoning by analogy: (1) given the initial conception A, the analoguous conception B must "come to mind"; (2) the analogy relation must be "confirmed"; and (3) conception B must be "confirmed." The paper also identifies three major questions to be considered in future research: (1) how do analogous cases come to mind? (2) how are the analogy relations and new conceptions confirmed or disconfirmed? and (3) why do they have explanatory power? (TW)